Integration of Three-Phase Microelectroextraction Sample Preparation Into Capillary Electrophoresis

Journal of Chromatography A 1610 (2020) 460570

Contents lists available at ScienceDirect

Journal of Chromatography A

journal homepage: www.elsevier.com/locate/chroma

Integration of three-phase microelectroextraction sample preparation into capillary electrophoresis

∗ Amar Oedit a, Bastiaan Duivelshof a, Peter W. Lindenburg a,b, , Thomas Hankemeier a a Division of Systems Biomedicine and Pharmacology, Leiden Academic Centre for Drug Research, Einsteinweg 55, 2300 RA Leiden, the Netherlands b University of Applied Sciences Leiden, Faculty Science & Technology, Research Group Metabolomics, Mailbox 382, 2300 AJ, Leiden, the Netherlands

a r t i c l e i n f o a b s t r a c t

Article history: A major strength of capillary electrophoresis (CE) is its ability to inject small sample volumes. However, Received 11 July 2019 there is a great mismatch between injection volume (typically < 100 nL) and sample volumes (typically

Revised 25 September 2019 20–1500 μL). Electromigration-based sample preparation methods are based on similar principles as CE. Accepted 25 September 2019 The combination of these methods with capillary electrophoresis could tackle obstacles in the analysis of Available online 26 September 2019 dilute samples. Keywords: This study demonstrates coupling of three-phase microelectroextraction (3PEE) to CE for sample Electromembrane extraction preparation and preconcentration of large volume samples while requiring minimal adaptation of CE Electroextraction equipment. In this set-up, electroextraction takes place from an aqueous phase, through an organic ﬁlter Sample preparation phase, into an aqueous droplet that is hanging at the capillary inlet. The ﬁrst visual proof-of-concept for

Capillary electrophoresis this set-up showed successful extraction using the cationic dye crystal violet (CV). The potential of 3PEE Sample enrichment for bioanalysis was demonstrated by successful extraction of the biogenic amines serotonin (5-HT), tyro- Electrophoresis sine (Tyr) and tryptophan (Trp). Under optimized conditions limits of detection (LOD) were 15 nM and 33 nM for 5-HT and Tyr respectively (with Trp as an internal standard). These LODs are comparable to other similar preconcentration methods that have been reported in conjunction with CE. Good linearity

( R 2 > 0.9967) was observed for both model analytes. RSDs for peak areas in technical replicates, interday and intraday variability were all satisfactory, i.e., below 14%. 5-HT, Tyr and Trp spiked to human urine were successfully extracted and separated. These results underline the great potential of 3PEE as an integrated enrichment technique from biological samples and subsequent sensitive metabolomics analysis. ©2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license. ( http://creativecommons.org/licenses/by-nc-nd/4.0/ )

1. Introduction acceptor phase and thereby leads to stacking and preconcentration of analytes. Electromigration-based techniques offer several Sample preparation is a crucial aspect of bioanalysis. The main advantages over partitioning-based techniques, such as being able objectives of sample preparation are to purify and enrich analytes to handle small sample volumes, enhanced extraction speeds (due prior to separation and detection. Commonly used sample prepa- to the electric field being the driving force rather than partitioning ration techniques are protein precipitation and partitioning based between phases) and ease of automation [4–6] . techniques, e.g., solid phase extraction (SPE) and liquid–liquid The combination of electromigration-based sample pretreat- extraction [1,2] . In the past years, electromigration-based extrac- ment with CE offers two main benefits. First, both approaches are tion methods have gained increased attention [3–5] . The main based on electromigration, so compounds that can be extracted are principle behind electromigration-based techniques is the use of also suited for CE separation. Second, electromigration-based tech- an electric field to extract ions from a donor phase (optionally niques can help overcome one of the drawbacks of CE: there is a through intermediate phases) to an acceptor phase. The migration great mismatch between the injected volume (typically < 100 nL) speed depends on the electrophoretic mobility of the analyte and and the sample volume (typically 20–1500 μL). In order to over- the electric field. The electric field strength is typically low in the come this mismatch miniaturized inserts have been developed [7] . However, when samples are too dilute and compounds fall below detection limits this does not provide a solution. Electromigration- ∗ Corresponding author. based sample preparation can help overcome the mismatch that E-mail addresses: [email protected] , [email protected]

(P.W. Lindenburg). is often present between injection volumes and sample volumes https://doi.org/10.1016/j.chroma.2019.460570 0021-9673/© 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license. ( http://creativecommons.org/licenses/by-nc-nd/4.0/ ) 2 A. Oedit, B. Duivelshof and P.W. Lindenburg et al. / Journal of Chromatography A 1610 (2020) 460570 in CE analysis and can offer significant advantages in loadability Previously, we have reported 3PEE as a powerful preconcentra- compared to in-line stacking methods. On-line SPE-CE, where SPE tion technique. Herein, we propose integration of this system into is coupled to the capillary, could also serve as a solution for en- a CE set-up for on-line sample clean-up and preconcentration. Ra- hancing the loadability, but has thus far only been demonstrated terink et al. demonstrated the extraction of acylcarnitines from a for more apolar compounds on apolar cartridges [8] . Moreover, to 50 μL donor phase, through a filter phase into a 2 μL acceptor the best of our knowledge, no commercial solutions for SPE-CE are phase that was formed by a conductive pipet tip prior to direct available and setting them up is complex. analysis with high-resolution mass spectrometry. In 3PEE, altering In electromembrane extraction (EME) a supported liquid mem- the organic filter phase composition influences the selectivity and brane (SLM) is used to separate the donor from the acceptor phase. enables selective extraction of desired analytes [21] . The analytes are extracted through the SLM into the acceptor In this article, we demonstrate a proof-of-principle of a novel phase using an electrical potential. Several set-ups in which EME on-line analytical system in which electromigration-based sample was coupled off-line to CE-UV and applied to bioanalysis have been preparation technique, i.e., 3PEE, is directly hyphenated to CE-UV. reported [6,9–15] . Our method offers several advantages: automation, sample precon- Off-line, at-line, as well as in-line coupling of EME to CE-UV centration and the ability to extract analytes from salt-rich ma- has been reported on several occasions. An off-line process con- trices without sample preparation. In order to enable 3PEE the sists of two steps (e.g. sample preparation followed by separation). electrode configuration of a commercially available CE apparatus An at-line process combines the two steps via an automated was modified. This modification entails placing an insulating sleeve handler (e.g. a robot). An in-line process takes place within the over the electrode whilst leaving the bottom part of the electrode separation system (e.g. SPE-CE with sorbent inside the capillary). exposed. 3PEE-CE-UV permits automated on-line selective analyte An on-line process takes place right before the sample stream extraction and enrichment, directly from a dilute sample. In this enters the separation system, but does take place within the set-up the analytes are extracted from a sample vial containing a analytical instrument (e.g. SPE-CE with sorbent outside the cap- two-phase liquid-liquid system, i.e., the aqueous sample and an illary) [16] . An example of on-line coupling of EME to CE-UV, is organic filter phase. Extraction takes place into a droplet of ac- nano-EME, for preconcentration and analysis of drugs. Here, basic ceptor phase hanging at the capillary inlet in the organic filter drugs were extracted from a sample volume of 200 μL, through a phase. membrane over a cracked capillary, into an acceptor of ∼8 nL. A In a first series of experiments the process was visualized us- single SLM could last for more than 200 extractions. Enrichment ing the cationic dye CV to assess the stability of the droplet of ac- factors (EFs) ranging between 25 and 196 were reported for basic ceptor phase during 3PEE. Then, the proposed method was eval- drug compounds, corresponding to recoveries of 0.1% and 0.79%. uated using selected biogenic amine model compounds to evalu- Under different conditions, higher EFs were obtained, up to 500 ate its potential for bioanalysis of polar metabolites. In order to for loperamide [17] . Moreover, EME-CE for preconcentration and enhance the CE separation, pH-mediated stacking was included analysis of basic drugs was reported. Here, an SLM was formed in the system. Consecutively, important extraction parameters (EE between two conical polypropylene units and extraction took voltage and EE time) and selectivity of the developed 3PEE-CE- place from the device, which was used as a vial insert. Extrac- UV method were investigated to optimize extraction. Then, the tion took place from a 40 μL donor compartment over the SLM analytical performance of 3PEE-CE-UV was compared to conven- into an acceptor compartment of 40 μL and therefore it only tional CE-UV. Finally, the performance of 3PEE-CE-UV as a sam- served for sample cleanup and not for sample preconcentration. ple preparation procedure for polar compounds in salt-rich bi- Recoveries ranging from 37% to 84% were obtained [18] . However, ological samples was investigated by analysis of spiked human the device requires reassembly for each new experiment, which urine samples in order to show its applicability for metabolomics hampers automated analysis. Chui et al. integrated the free liquid analyses. membrane (FLM) in an electrokinetic supercharging (EKS) method in-line to further improve detection limits in CE. In this approach, 2. Materials and methods a small plug of immiscible organic solvent in the capillary was used as filter during the electrokinetic sample injection to enhance 2.1. Chemicals and reagents stacking efficiency. Analysis of cationic herbicides in environmental water samples were used to evaluate the on-line preconcentra- Sodium chloride (NaCl), CV, ammonium hydroxide ( > 25%), 5- tion efficiency and results showed detection limit enhancements HT, Tyr and Trp were obtained from Sigma-Aldrich (Steinheim, Ger- of over 1500 times. EKS over an FLM using CE for analysis on many). Ethyl acetate (EtOAc) was obtained from Actu-All (Oss, The real-world samples has only been demonstrated on river water Netherlands). Formic acid (FA) was obtained from Acros Organics samples, a relatively clean matrix. Since the donor phase is drawn (Geel, Belgium). Sodium hydroxide was obtained from VWR (Am- partially into the capillary, protein-rich samples could cause sterdam, The Netherlands). All solutions were of HPLC grade or reproducibility problems due to absorption to the fused silica higher. Water was prepared using a Milli-Q R Advantage A10 R sys- capillaries [19] . tem (Billerica, MA, USA). EME can be used without SLM. This offers several advantages, such as not requiring reassembly and regeneration of the SLM. In 2.2. Samples and stock solutions this case the technique is referred to as EME over an FLM [20] or 3PEE [21] . Off-line coupling of a EME-FLM extraction to CE-UV was Aqueous stock solutions of analytes (2 mM) were stored at − successfully demonstrated for analysis of charged basic drugs (nor- 20 °C until use. Sample solutions were prepared at the desired triptyline, haloperidol and loperamide) in human urine and blood concentration by dissolving the aqueous stock solutions in 1 M FA [22] . (pH 1.8) and were, at maximum, kept at 4 °C for 1 week. Human − Two-phase electroextraction (EE) is a process where charged urine was provided by a healthy volunteer and stored at 20 °C. analytes are transferred from an organic donor phase into an aque- Prior to analysis, urine was thawed and centrifuged for 15 min at ous acceptor phase using an electric field [23–25] . EE coupled to 10,0 0 0 rpm, the urine was spiked with the model analytes in the CE was first reported in conjunction with isotachophoresis by Van desired concentration and FA was added until 1 M was reached in der Vlis et al. in 1994 [26] . urine. A. Oedit, B. Duivelshof and P.W. Lindenburg et al. / Journal of Chromatography A 1610 (2020) 460570 3

2.3. Equipment and techniques the donor phase through the FLM into the acceptor droplet. The septa of the inlet vials were removed to ensure that the modiﬁed 2.3.1. Capillary electrophoresis electrode, which had a slightly increased thickness due to the PTFE Analyses were performed using a Beckman Coulter P/ACE MDQ sleeve, could still reliably enter the vial. (Fullerton, CA, USA) CE apparatus using UV diode array detection. In order to visually monitor the extraction process a USB-pen A fused silica capillary of 50 μm I.D. and 365 μm O.D. with a to- video camera was mounted inside the CE machine and focused on tal length of 60 cm was used (Polymicro Technologies, USA). New the capillary inlet. Debut Video Capture (NCH Software, Greenwood capillaries were sequentially rinsed at 1379 mbar with MeOH for Village, CO, USA) was used to record the extraction videos. 10 min, 1 M NaOH for 10 min, water for 5 min and background elec- trolyte (BGE) for 20 min. Between runs, the capillaries were ﬂushed 3.2. Three-phase electroextraction procedure for 5 min with BGE.

Separation was performed using 1 M FA (pH 1.8) as the BGE Prior to placing the sample vials in the CE system, 375 μL donor + buffer using a separation voltage of 17.5 kV. The capillary car- solution was pipetted in conventional CE vials (1.5 mL). Based on tridge temperature was set at 20 °C. Detection was set at 195 nm previous EE works the donor solution was acidified to 1 M FA, to maximize the number and response of metabolites that can be which has proven to be a good donor solvent [23–25] . This was fol- detected with a reference at 400 nm. lowed by 725 μL organic filter phase consisting of water-saturated EtOAc, which is crucial for the electric field and thereby the trans- 2.3.2. Software fer of ions through the organic filter layer [21] . Fig. 1 graphically 32 Karat (Beckman Instruments, Fullerton, CA, USA) was used depicts a typical 3PEE experiment in the CE instrument. First, the for controlling the CE-UV system and for data acquisition. Injection capillary was rinsed with FA. Then ammonium hydroxide solution volumes as well as the volumes of the acceptor droplet formed was injected, followed by an injection of BGE. After inserting the by reversed pressure were calculated using Sciex CE Expert V2.2 capillary in the sample vial, a hanging droplet of 100 nL is created (Framingham, MA, USA). in the organic filter phase by applying a pressure of −69 mbar for 1 min from the BGE outlet vial. Then, electroextraction was per-

3. Results and discussion formed by applying the extraction voltage, after which the enriched droplet was retracted into the capillary. At last, the capillary

3.1. Modification of the CE instrument for three-phase inlet was inserted into a BGE vial and separation was carried out. electroextraction In order to visualize the electroextraction procedure, the cationic dye CV was added to the donor phase. In order to enable 3PEE using a Beckman Coulter CE apparatus the electrode configuration was modified by replacing the ex- 3.3. Visualization of on-line three-phase electroextraction isting electrode with a longer platinum electrode of 4 cm ( Fig. 1 B). From the bottom 2.8 cm of the electrode was isolated using a poly- The setup for 3PEE hyphenated to CE-UV was based on the tetrafluoroethene (PTFE) sleeve, leaving only a tip of 2 mm of the previously reported 3PEE-DI-MS [21] . In a visual proof-of-concept electrode exposed. This modification enables an electric field from 10 μM CV was electroextracted at 3.5 kV from 375 μL donor phase

Fig. 1. (a) Schematic representation of the 3PEE setup, and (b) actual set-up incorporating the modiﬁed electrode conﬁguration used during experiments (bottom of vial not visible). (c) Schematic representation of the key steps in the extraction procedure in the CE-UV system: (1) injection of ammonium hydroxide, (2) injection of BGE, (3) application of negative pressure, (4) application of voltage, (5) retraction of droplet using pressure, (6) vial switch to BGE and start of CE separation. 4 A. Oedit, B. Duivelshof and P.W. Lindenburg et al. / Journal of Chromatography A 1610 (2020) 460570

Tyr, Trp and 5-HT compared to lower voltages ( Fig. 4 B and C). Volt- ages beyond 3 kV resulted in loss of current and droplet instability (data not shown). Subsequently, the extraction time was optimized while keeping the extraction voltage constant at the optimal value of 3 kV. Extrac- tions were performed for 2, 4, 6, 8 and 10 min. It was shown that increasing the extraction time increased enrichment and thereby peak areas of the analytes. Beyond 8 min of extraction caused fre-

Fig. 2. Visualization of 3PEE coupled CE using CV. Video stills of: (A) initial condi- quent current losses during CE. tions with sample vial containing 375 μL crystal violet (10 μM) and 725 μL water- saturated ethyl acetate; (B) end of droplet formation ( ∼100 nL, 1 M FA), outline of In summary, the optimized 3PEE procedure was as fol- droplet barely visible; (C) start 3PEE by application of 3 kV; (D) end of 3PEE after lows. First, the capillary was flushed with 1 M FA for 5 min 10 min. at 1378 mbar, followed by a 17 s 34 mbar injection of 15% ammonium hydroxide and subsequent 1.1 min 69 mbar injection of 1 M FA. Then, a droplet was formed using 1 min into ∼100 nL acceptor phase ( Fig. 2 A–C). The liquid-liquid interface 69 mbar, after which electroextraction was carried out at 3 kV for is not visible in the sample tray (see Supporting Information S1 8 min. Finally, the enriched droplet was retracted using 1.5 min for vial outside of sample tray). Prior to application of the extrac- 34 mbar and CE-UV separation was performed for 35 min at tion voltage, no CV can be observed in the droplet. This indicates 17.5 kV. that the contribution of partitioning to the process is minimal. Af- ter 10 min electroextraction, the droplet was enriched dramatically with CV ( Fig. 2 D). 3.4.3. Analytical figures of merit These results indicate that the developed set-up is successfully Table 1 shows the analytical performance of the optimized extracting CV from the donor phase into the pendant acceptor method for the biogenic amines 5-HT and Tyr in comparison with phase. conventional CE-UV. Trp was used as internal standard. The aforementioned conditions were used to evaluate the ex- 3.4. On-line three-phase electroextraction coupled to capillary traction of 5-HT and Tyr using different concentrations (0, 0.05, 0.1, = electrophoresis 0.5, 1, 5 μM; n 3) resulting in a linear range of 0.01–5 μM, yield- ing regression coefficients ( R 2 ) of 0.9967 and 0.9995, respectively 3.4.1. Effect of pH-mediated stacking ( Table 1 ). LODs were estimated using signal-to-noise (S/N) ratios = The 3PEE-CE-UV was investigated using the biogenic amines of triplicate measurements at 50 nM and extrapolated to S/ N 3. Tyr, Trp and 5-HT. The BGE consisted of 1 M FA (pH 1.8) to ensure Detection limits of 15 nM and 33 nM were observed for 5-HT and compounds were cationic during analysis. In a first experiment, Tyr, respectively. 500 nM model analytes were extracted for 6 min at 3 kV and the For comparison, calibration curves were constructed with droplet was partially retracted at 34 mbar for 5 s, followed by CE identical electrophoresis conditions using hydrodynamic injection separation at 17.5 kV for 30 min. In Fig. 3 A it can be observed that without pH-mediated stacking injecting 80 nL (similar to the re- Trp and Tyr have poorly resolved peaks, with 5-HT overlapping. tracted volume using the optimized 3PEE-CE-UV method) using = The peak areas are higher despite reduced injection time compared different concentrations (0, 1, 5, 10, 25, 50 μM; n 3) of 5-HT to Fig. 3 B. This is caused by migration of analytes from the donor and Tyr with 25 μM Trp as internal standard. Since CE-UV could phase through the droplet into the capillary (see Supporting In- not reach the nM range of 3PEE-CE-UV the examined range was formation S2). This can possibly be explained by peak broaden- adjusted to a micromolar range to be able to construct a cali- ing during 3PEE caused by electrophoresis and EOF, which is still bration curve ( Table 1 ). Regression analysis yielded high R ² val- > present to some extent, even at a low pH. In order to focus the ues (exceeding 0.999) for 5-HT and Tyr with conventional CE- broad sample zone and thereby improve separation, a pH-mediated UV and observed LODs for 5-HT and Tyr were 5 μM and 1 μM, re- stacking was included. By adding a plug of basic BGE after to the spectively. The LODs for the 3PEE-CE-UV method were improved ∼ × ∼ × acceptor phase, the acidic BGE titrates the sample solution to cre- 333 and 30 for 5-HT and Tyr, respectively. Moreover, com- ate a neutral zone. In this zone a higher field is present causing pared to conventional CE-UV, linear range of 3PEE-CE-UV was increased migration speed of analytes and eventually stacking at extended an order of magnitude downwards to the 50–100 nM the interface between the neutral zone and BGE. This pH-mediated range. stacking was created by injecting 15% ammonium hydroxide solution for 17 s at 34 mbar and followed by 1 M FA for 1.1 min at 3.4.4. Repeatability and technical replicates 69 mbar to ensure that the droplet consisted fully of BGE. The Intra- and inter-day variability of the method were determined biogenic amines were extracted at a concentration of 500 nM for using optimized conditions at 500 nM. As shown in Table 2 , intra- 8 min at 3 kV. The droplet could now be retracted much longer day variability analysis showed good repeatability, as RSDs for AUC (1.5 min at 34 mbar) while the separation resolution improved as values ranged from 4.7% for Tyr up to 6.9% for 5-HT. For inter-day shown in Fig. 3 B. variability, RSD values ranged between 7.9% for Tyr and 13.8% for 5-HT, indicating good repeatability of the developed method. The 3.4.2. Optimization of extraction voltage and extraction time increased RSD values obtained compared to conventional CE can The method was optimized in order to obtain the highest pos- be partially explained by the added steps, including droplet forma- sible area under the curve (AUC) for the analytes. In the first se- tion and extraction vs. hydrodynamic injection. ries of experiments ( n = 3) the extraction time was kept constant Technical replicates of a single sample vial showed similar vari- at 5 min and the extraction voltage was varied (1, 1.5, 2, 2.5, 3, 4 ability. It was shown that five consecutive extractions could be per- and 5 kV). In these experiments 250 nM Trp, Tyr and 5-HT were formed successfully and resulted in RSD values of 6.6% for 5-HT used. When 0 kV was used no analytes were detected ( Fig. 4 A). and 10% for Tyr, comparable to the 3PEE-CE-UV inter- and intra- This indicates that analyte migration from the donor phase to the day variability ( Table 2 ). Migration time repeatability of the new acceptor phase is solely driven by electric potential. Moreover, in- method was comparable to hydrodynamic CE (Supporting Informa- creasing the voltage up to 3 kV significantly enhanced signals for tion S4). A. Oedit, B. Duivelshof and P.W. Lindenburg et al. / Journal of Chromatography A 1610 (2020) 460570 5

Fig. 3. Stacking effects in 3PEE coupled to CE-UV. Top ﬁgures show the current proﬁle of extraction of 500 nM 5-HT (1), Trp (2) and Tyr (3) from 375 μl donor phase using 3PEE-CE-UV as well as the corresponding electropherograms. (A) retracting for 0.5 min at 6.9 mbar and (B) retracting for 1.5 min at 34 mbar with pH-mediated stacking.

AB2

1 3

60000 50000 50000 40000 40000 alytes n 30000 30000

20000 20000 AUC of a 10000 UC of analytes A 10000

0 0 1 .5 2 .5 3 1 2 2 4 6 8 Extraction voltage (kV) Extraction time (min)

Fig. 4. 3PEE voltage and time optimization. Extractions were performed using 250 nM 5-HT (1), Trp (2) and Tyr (3) in 1 M FA. Electropherograms are shown for: (A) 0 kV 3PEE voltage and (B) 3 kV 3PEE voltage. (C) Results of voltage and time optimization for 3PEE coupled online to CE-UV. Error bars represent standard deviation of n = 3 . 6 A. Oedit, B. Duivelshof and P.W. Lindenburg et al. / Journal of Chromatography A 1610 (2020) 460570

Table 1 Comparison the analytical performance of 3PEE-CE-UV and conventional CE-UV in neat solutions. AUCs were corrected using Trp as internal standard.

− Analyte Linear range (μM) Sensitivity ± SD ( ×10 −2 AU/μM) Intercept ± 95% CI a ( ×10 2 AU/μM) Linearity ( R ²) LOD b (μM)

CE 3PEE CE 3PEE CE 3PEE CE 3PEE CE 3PEE

5-HT 5–50 0.05–5 2.54 ± 0.01 102.2 ± 1.3 −0.40 ( −2.5–2.5) −2.10 ( −11.6–7.4) > 0.9999 0.9995 5 0.015 Tyr 1–50 0.1–5 9.65 ± 0.06 76.8 ± 2.6 1.67 ( −3.5–6.9) 5.98 ( −12.8–24.8) 0.9999 0.9967 1 0.033

Note: for repeatability see Table 2 .

a No signiﬁcant intercept values were observed ( p < 0.05).

b Extrapolation towards S/N of 3 from lowest measured concentration.

Table 2 Intra- and interday repeatability and technical replicates of 3PEE-CE-UV analysis of target compounds. AUCs were corrected using Trp as internal standard.

Analyte 3PEE-CE-UV intraday ( n = 3 ) CE-UV intraday ( n = 3) 3PEE-CE-UV interday ( n = 6) 3PEE-CE-UV technical replicates a ( n = 5)

Mean AUC ratio RSD area ratio (%) Mean AUC ratio RSD area ratio (%) Mean AUC ratio RSD area ratio (%) Mean AUC ratio RSD area ratio (%)

5-HT 0.34 6.9 0.12 1.85 0.35 13.8 0.41 6.6 Tyr 0.44 4.7 0.50 0.87 0.43 7.9 0.43 10.0

a Obtained from 5 consecutive extractions from a single sample vial.

3.4.5. Enrichment and recovery netic supercharging over FLM set-up incorporates a t-ITP step to In order to assess the performance of 3PEE-CE-UV correctly, the further enhance stacking of the analytes rather than pH-mediated extraction recovery (ER) and EF were calculated [6] . stacking. Most preconcentration set-ups in Table 3 are coupled to These results show that even though the EF max is much greater, CE are reported for analysis of apolar basic drug compounds, with enrichment was limited. Tyr in the acceptor phase was around 8 the exception electrokinetic supercharging over an FLM [19] , which times more concentrated than the donor phase after extraction was used polar herbicides. Unlike electrokinetic supercharging over (Supporting information S3). It was observed that the enrichment an FLM, 3PEE-CE-UV does not require removal of FLM from the factor of 5-HT was 41.4 and therefore five times higher than for capillary, which can be a convoluted procedure and requires re- Tyr. A possible explanation for this is the lack of the carboxylic optimization for each new organic FLM phase. acid moiety (p K a = 2.38) in 5-HT, thereby enabling more efficient 3PEE-CE-UV has a relatively long total analysis time compared transfer from the FA containing donor phase. Likewise, a lower re- to the discussed set-ups. However, as this is a proof-of-principle, covery was observed for Tyr (0.2%) than for 5-HT (1.1%) after ex- separation parameters such as separation voltage and capillary traction which correlates with the EF values. The improvements in length could still be optimized to yield shorter analysis times. The LOD in Table 1 differ from the obtained EF values as the devel- obtained EFs of 3PEE-CE-UV were 1–2 orders of magnitude lower oped method included both stacking through on-line electroextrac- than other reported methods. However, due to the combination tion (increasing loading) and in-line stacking through a dynamic with dynamic pH-mediated stacking similar LODs were obtained. pH-mediated stacking (improving peak shapes). Both techniques A probable explanation for this is the fact that we studied the po- are essential to the final method and therefore the final method tential of our method for polar metabolites, while in other work was compared to a simple hydrodynamic injection method (thus apolar drugs, which are more easily extracted are studied. Transfer without dynamic pH-mediated stacking). These results show that of polar molecules, such as the biogenic amines in this work, has the extraction process is not exhaustive and is a soft extraction always been a challenging endeavor in EME and tuning of com- method, which offers several advantages such as opening up the position and size of the organic phase remains at the forefront possibility of studying (bio)chemical reactions and concentration- of interest [10] . The developed 3PEE-CE-UV has LODs in the low time monitoring without disturbing the overall system. EF and ER nM range, which is similar to other techniques in Table 2 , despite can be further improved by reducing the volume (and thereby having lower EFs likely due to its higher injection volume com- height) of the organic phase to enhance the electric field distri- bined with in-capillary stacking. The low recoveries of 3PEE-CE-UV bution to be more favorable towards analyte extraction. Moreover, make it suitable as a soft extraction technique. Moreover, on-line the composition of the organic phase can be modified to enhance methods such as described in Table 3 are more suited for analy- EF and ER as well [21] . sis of large dilute samples as these can handle larger sample sizes compared to high nL range samples in in-line methods. Finally, as

3.4.6. Comparison to other set-ups 3PEE-CE-UV is not exhaustive it can be used to measure technical

A comparison of 3PEE-CE-UV method to other sample extrac- replicates (i.e. repeat analyses of the same sample vial). tion techniques that were hyphenated directly to CE and reported in literature is shown in Table 3 and Supporting Information S5. 3.4.7. Proof of concept: urine bioanalysis Single drop micro-extraction (SDME) techniques have been cou- The potential of on-line 3PEE as sample cleanup procedure for pled to CE [27,28] with EFs ranging between 130–150 and EME has metabolomics analyses was investigated by analyzing human urine been coupled on-line to capillaries [17] with reported EFs rang- with 5-HT, Trp and Tyr spiked to it. Analyte stocks were dissolved ing between 25–196, with loperamide reaching an EF of up to 500 in urine and acidified to 1 M FA (pH 1.8) prior to electroextrac- under optimal conditions. On-line back extraction field amplified tion. Field amplified stacking is reduced when analytes are dis- sample injection relies on both partitioning between an organic solved in a highly conductive matrix and effective dynamic pH- chloroform donor phase and an aqueous acceptor phase and simul- mediated stacking requires a shorter plug of ammonium hydroxide taneous depletion of the acceptor phase via electrokinetic injection for optimal stacking [30] . As urine is a highly conductive matrix into the capillary [29] . 3PEE-CE-UV bears similarities to the elec- the length of the ammonium hydroxide plug was reduced (8 s at trokinetic supercharging over FLM set-up, but differs in two ways: 34 mbar). In order to further increase the stacking efficiency, the (1) preconcentration takes place at the capillary inlet into a pen- droplet retraction time was reduced (68 s at 34 mbar). This shorter dant droplet rather than inside the capillary, and (2) the electroki- retraction resulted in an improved stability and comparable extrac- A. Oedit, B. Duivelshof and P.W. Lindenburg et al. / Journal of Chromatography A 1610 (2020) 460570 7

Table 3 Comparison of the newly developed 3PEE-CE-UV method to other methods that were hyphenated to CE.

Total analysis Set-up Ref. Compounds EF (range) ER (range) LOD (range) time (min)

3PEE-CE-UV This paper Serotonin , Tyrosine , 7.8 –42 0.2 –1.1 % 15 –33 nM 52 (Phenylalanine) Inline SDME-CE-MS [27] Methoxyphenamine , 130 –150 Not 2 –5 nM 62 Methamphetamine , reported Amphetamine, Phenetylamine

Nano-EME coupled to [17] Pethidine , 25 –196 0.1% −0.79 % 8 –31 nM (0.2 – > 29 d

CE-UV Nortiptyline, 15 ng mL −1) Methadone, Haloperidol, Loperamide

FLM – electrokinetic [19] Paraquat, Diquat Not (Relative 58 –58 nM > 20 d

supercharging coupled reported recovery a : (0.15–

to CE-UV 97.0– 0.20 ng mL −1)

97.5% b ) On-line back [28] Cocaine, Thebaine, Not ( Relative 16 –16 nM 18

extraction ﬁeld (Metamphetamine) reported recovery a : (0.005–

ampliﬁed sample 94.71– 0.005 μg mL −1)

injection (coupled to 98.65% c ): CE-UV) Underlined compounds and values are those with the lowest LOD, compounds in italics are those with the highest LOD.

a Relative recoveries were reported by measuring a spiked sample and comparing to the calibration curve. − b Spiked at 20 ng mL 1. − c Spiked at 0.5 μg mL 1.

d Duration of initial ﬂush prior to each analysis not speciﬁed.

Fig. 5. Proof of concept showing urine bioanalysis using 3PEE-CE-UV. Electropherograms obtained from (A) non-spiked urine and (B) spiked urine samples extracted by 3PEE prior to CE-UV detection. Urine was spiked with 5-HT (1), Tyr (3) and Trp (2; 50 μM). Extraction and analytical conditions can be found in Section 3.6. tion and separation current profiles to extractions performed in tive detector is required. These preliminary results show the poten- neat solutions (data not shown). Before analysis of the target ana- tial of 3PEE-CE-UV as an easy on-line sample preparation method lytes, 3PEE-CE-UV was first applied to a non-spiked urine sample. that could be applied for the analysis of small polar metabolites, In the corresponding electropherogram at the non-specific wave- e.g., in a metabolomics setting. length 195 nm, many unidentified endogenous compounds were observed after the extraction procedure showing its ability to ana- 4. Conclusion lyze a urine sample without requiring prior dilution ( Fig. 5 A). Then, 3PEE-CE-UV was employed for the analysis of 5-HT, Trp and Tyr In this work we have successfully developed a new approach (50 μM), spiked to urine of the same origin. The electropherogram for simultaneous sample preconcentration and cleanup. This was in Fig. 5 B, shows detection of 5-HT, Trp and Tyr. In order to confirm achieved through integration of 3PEE with CE-UV, which required the identities of the endogenous compounds in urine a more selec- a simple modification in the electrode configuration of a commer- 8 A. Oedit, B. Duivelshof and P.W. Lindenburg et al. / Journal of Chromatography A 1610 (2020) 460570 cially available CE instrument. By placing an immiscible organic [9] A. Šlampová, P. Kubánˇ , Two-phase micro-electromembrane extraction across filter phase on top of an aqueous sample, cationic analytes were free liquid membrane for determination of acidic drugs in complex samples, Anal. Chim. Acta 1048 (2019) 58–65 . extracted into an aqueous acceptor droplet formed at the capil- [10] N. Drouin , S. Rudaz , J. Schappler , New supported liquid membrane for elec- lary inlet, when an electrical field was applied. In order to enable tromembrane extraction of polar basic endogenous metabolites, J. Pharm. full droplet retraction without loss of resolution after extraction, Biomed. Anal. 159 (2018) 53–59 . ˇ [11] M. Silva, C. Mendiguchía, C. Moreno, P. Kubán, Electromembrane extraction and pH-mediated stacking was introduced to efficiently stack analytes. capillary electrophoresis with capacitively coupled contactless conductivity de- The performance of on-line 3PEE-CE-UV was evaluated by extract- tection: multi-extraction capabilities to analyses trace metals from saline sam- ing 5-HT, Trp and Tyr from a 375 μL neat solution into a pendant ples, Electrophoresis 39 (2018) 2152–2159 . droplet of 100 nL BGE. Low extraction recoveries were obtained, [12] M. Dvorákˇ , K.F. Seip, S. Pedersen-Bjergaard, P. Kubánˇ, Semi-automated set-up for exhaustive micro-electromembrane extractions of basic drugs from biolog- demonstrating that the technique is a soft extraction technique. To ical fluids, Anal. Chim. Acta 1005 (2018) 34–42 . the best of our knowledge, EME over an FLM of polar metabolites [13] A. Šlampová, P. Kubánˇ , Direct analysis of free aqueous and organic operational was never reported. It was demonstrated that detection limits im- solutions as a tool for understanding fundamental principles of electromembrane extraction, Anal. Chem. 89 (2017) 12960–12967 . proved to 15 nM 5-HT and 33 nM Tyr with 3PEE, compared to 5 μM [14] M. Forough , K. Farhadi , A. Eyshi , R. Molaei , H. Khalili , V. Javan Kouzegaran , 5-HT and 1 μM Tyr in CE with hydrodynamic sample injection. As A .A . Matin , Rapid ionic liquid-supported nano-hybrid composite reinforced proof-of-concept, the on-line 3PEE-CE-UV procedure was evaluated hollow-fiber electromembrane extraction followed by field-amplified sample injection-capillary electrophoresis: an effective approach for extraction and for the analysis of human urine. It was demonstrated that 5-HT, quantification of imatinib mesylate in human plasma, J. Chromatogr. A 1516 Trp and Tyr were successfully extracted from spiked urine, thus (2017) 21–34 . signifying the potential of the developed procedure for urine bio- [15] A.R. Fakhari , H. Mohammadi Kosalar , S. Asadi , K.S. Hasheminasab , Surfac- tant-assisted electromembrane extraction combined with cyclodextrin-modi- analysis and metabolomics. fied capillary electrophoresis for the separation and quantification of tranyl- cypromine enantiomers in biological samples, J. Sep. Sci. 41 (2018) 475–482 . Declaration of Competing Interest [16] R. Ramautar , G.W. Somsen , G.J. de Jong , Recent developments in coupled SPE-CE, Electrophoresis 31 (2010) 44–54 . [17] M.D. Payan , B. Li , N.J. Petersen , H. Jensen , S.H. Hansen , S. Pedersen-Bjergaard , None. Nano-electromembrane extraction, Anal. Chim. Acta 785 (2013) 60–66 . [18] P. Pant u˚ cková,ˇ P. Kubánˇ , P. Bo cekˇ , A simple sample pretreatment device with

Acknowledgements supported liquid membrane for direct injection of untreated body ﬂuids and in-line coupling to a commercial CE instrument, Electrophoresis 34 (2013) 289–296 . The research leading to these results has received funding [19] M.Q. Chui , L.Y. Thang , H.H. See , Integration of the free liquid membrane into from the European Community’s Seventh Framework Program electrokinetic supercharging - capillary electrophoresis for the determination of cationic herbicides in environmental water samples, J. Chromatogr. A 1481

(FP7/2007-2013) under Grant Agreement No. 306000 (STATegra) (2017) 145–151 . and Grant Agreement No. 602783 (CAM-PaC). The authors would [20] P. Kubánˇ , P. Bo cekˇ , Micro-electromembrane extraction across free liquid mem- like to acknowledge Raphaël Zwier for his help in adjusting the branes. instrumentation and basic principles, J. Chromatogr. A 1346 (2014) 25–33 . electrode conﬁguration of the CE instrument. [21] R.J. Raterink , P.W. Lindenburg , R.J. Vreeken , T. Hankemeier , Three-phase electroextraction , a new (online) sample puriﬁcation and enrichment method for Supplementary materials bioanalysis, Anal. Chem. 85 (2013) 7762–7768 . [22] P. Kuban , P. Bocek , Micro-electromembrane extraction across free liquid mem- branes. extractions of basic drugs from undiluted biological samples, J. Chro- Supplementary material associated with this article can be matogr. A 1337 (2014) 32–39 . found, in the online version, at doi: 10.1016/j.chroma.2019.460570 . [23] P.W. Lindenburg , R. Seitzinger , F.W.A. Tempels , U.R. Tjaden , J. van der Greef , T. Hankemeier , Online capillary liquid–liquid electroextraction of peptides as

fast pre-concentration prior to LC–MS, Electrophoresis 31 (2010) 3903–3912. References [24] P.W. Lindenburg , F.W. Tempels , U.R. Tjaden , J. van der Greef , T. Hankemeier , On-line large-volume electroextraction coupled to liquid chromatography-mass [1] I. Kohler , J. Schappler , S. Rudaz , Microextraction techniques combined with spectrometry to improve detection of peptides, J. Chromatogr. A 1249 (2012) capillary electrophoresis in bioanalysis, Anal. Bioanal. Chem. 405 (2013) 17–24 . 125–141 . [25] P.W. Lindenburg , U.R. Tjaden , J. van der Greef , T. Hankemeier , Feasibility of [2] R.-J. Raterink , P.W. Lindenburg , R.J. Vreeken , R. Ramautar , T. Hankemeier , Re- electroextraction as versatile sample preconcentration for fast and sensitive cent developments in sample-pretreatment techniques for mass spectrome- analysis of urine metabolites, demonstrated on acylcarnitines, Electrophoresis try-based metabolomics, TrAC Trends in Anal. Chem. 61 (2014) 157–167 . 33 (2012) 2987–2995 . [3] A. Wuethrich , P.R. Haddad , J.P. Quirino , The electric field –An emerging driver [26] E. Van der Vlis , M. Mazereeuw , U.R. Tjaden , H. Irth , J. van der Greef , Combined in sample preparation, TrAC Trends in Anal. Chem. 80 (2016) 604–611 . liquid-liquid electroextraction and isotachophoresis as a fast on-line focusing [4] P.W. Lindenburg , R. Ramautar , T. Hankemeier , The potential of electrophoretic step in capillary electrophoresis, J. Chromatogr. A (1994) 333–341 . sample pretreatment techniques and new instrumentation for bioanalysis, [27] Z.A. AlOthman , M. Dawod , J. Kim , D.S. Chung , Single-drop microextraction as a with a focus on peptidomics and metabolomics, Bioanalysis 5 (2013) . powerful pretreatment tool for capillary electrophoresis: a review, Anal. Chim. [5] S. Pedersen-Bjergaard , C. Huang , A. Gjelstad , Electromembrane extraction-Re- Acta 739 (2012) 14–24 . cent trends and where to go, J. Pharm. Anal. 7 (2017) 141–147 . [28] J. Kim , K. Choi , D.S. Chung , In-line coupling of single-drop microextraction with [6] A. Oedit , R. Ramautar , T. Hankemeier , P.W. Lindenburg , Electroextraction and capillary electrophoresis-mass spectrometry, Anal. Bioanal. Chem. 407 (2015) electromembrane extraction: advances in hyphenation to analytical techniques, 8745–8752 . Electrophoresis 37 (2016) 1170–1186 . [29] H. Fang , Z. Zeng , L. Liu , D. Pang , On-line back-extraction field-amplified sample [7] A. Šlampová, P. Kubánˇ , Injections from sub-μL sample volumes in commercial injection method for directly analyzing cocaine and thebaine in the extractants capillary electrophoresis, Journal of Chromatography A 1497 (2017) 164–171 . by solvent microextraction, Anal. Chem. 78 (2006) 1257–1263 . [8] R. Ramautar , G.W. Somsen , G.J. de Jong , CE-MS for metabolomics: develop- [30] Y.H. Tak , G.W. Somsen , G.J. de Jong , Optimization of dynamic pH junction for ments and applications in the period 2014–2016, Electrophoresis 38 (2017) the sensitive determination of amino acids in urine by capillary electrophore- 190–202 . sis, Anal. Bioanal. Chem. 401 (2011) 3275–3281 .